SMAGL Social Networks
Nick Mikulski
3/1/2021
1 Pre-Release Proximity Plot
1.1 Inputting Datasets & Editting the dataframe
pre_socnet<- curl("https://raw.githubusercontent.com/nickmikulski/Spring2021/main/Pre-release_Social%20Proximity_csv.csv")
pre_socnet<- read.csv(pre_socnet, header = T, na.strings=c(""," ","NA"))
#head(pre_socnet)
#str(pre_socnet)
pre_socnet_close<- pre_socnet %>% #creating a new dataframe called pre_socnet_close using data from the original pre_socnet dataframe
filter(Focal.ID != "BT", #this code REMOVES all data that has Batman has the focal ID (BT is wild male from prerelease)
Association != "BT", #this code REMOVES all data that has Batman in association column
Proximity.Code %in% c("1","2") #this code only keeps proximity codes 1,2 (excluding 3,4) because we are focusing on closer proximity
)
#head(pre_socnet_close)1.2 Creating a matrix
# 1. Create a character vector of all the focal IDs in dataset:
pre_sn_IDs<-as.character(unique(pre_socnet_close$Focal.ID))
#pre_sn_IDs
# 2. Get a list of dataframes, subsetted by monkey ID:
pre_sn_monkeylist<-lapply(pre_sn_IDs, function(x){pre_socnet[pre_socnet[["Focal.ID"]] == x, ]})
# The line above is a little bit confusing. It is creating a separate dataframe for each individual based on their focal id
#head(pre_sn_monkeylist)
# 3. Group each by focal/associate, and count how many times they are observed close together:
pre_sn_grouped<-
pre_sn_monkeylist %>%
purrr::map(~group_by(.,Association)) %>%
purrr::map(~summarize(.,count=n()))
#pre_sn_grouped
names(pre_sn_grouped) <- pre_sn_IDs #this gives each grouped list the name of the Focal ID
#pre_sn_grouped
# 4. Set up pairwise combinations of interacting monkeys:
pre_sn_monkeycombos<-list(focal=pre_sn_IDs, associate=pre_sn_IDs) #create list of all possible focals/associates
pre_sn_filtf<- function(x, y) {x == y} #create function to filter out same-monkey pairs ("PO is close to PO")
pre_sn_combo<- pre_sn_monkeycombos %>% cross_df(.,.filter=pre_sn_filtf) #get the filtered combined list as a dataframe
#pre_sn_combo1.3 Create the dataframe and matrix
# 5. Create new dataframes with specific criteria
pre_sn_combo2<-
pre_sn_combo %>%
mutate(absent1 = map2_chr( #new column called "absent1"
focal,
associate,
~if_else(.x %in% names(pre_sn_grouped),true="TRUE",false="FALSE"))) %>%
mutate(absent2 = map2_chr(
focal,
associate,
~if_else(.y %in% pre_sn_grouped[[.x]]$Association,true="TRUE",false="FALSE"))) %>%
filter(absent1 == "TRUE") %>%
filter(absent2 == "TRUE") %>%
dplyr::select(-absent1,-absent2) #this removes those two new columns you made so you're just left with the ID names
#Honestly I was quite confused on parts of this, but Laura/Dr. Schmitt gave me this helpful code
pre_sn_combo3<- pre_sn_combo2 %>%
mutate(proximity = map2_int( #new column called "proximity" that is the count for when proximity code = 1 or 2
focal,
associate,
~pre_sn_grouped %>% pluck(.x) %>% filter(Association==.y) %>% as.data.frame(.) %>% .[,2]))
#pre_sn_combo3
# 6. Create your matrix
pre_sn_matrix<-spread(pre_sn_combo3,associate,proximity) %>% column_to_rownames(var="focal") %>% data.matrix()
#pre_sn_matrix1.5 Plotting based on dataframe
My original intention was to use the matrix that was just created in order to plot the social network, however, I keep getting an error message that says "Some vertex names in edge list are not listed in vertex data frame. I tried trouble shooting this with no success, so I instead tried to work with the dataframe that created right before the matrix (pre_sn_combo3), and I had some success. Below is how I was originally trying to create the graph using the matrix.
Alternatives (?):
pre_sn_graph <- graph_from_data_frame(d=pre_sn_edges, vertices = pre_sn_nodes, directed = TRUE) #Using the matrix
sortproxf <- pre_sn_combo3[order(pre_sn_combo3$proximity),]
#sortproxf
pre_sn_edges <- pre_sn_df #assigning the matrix data frame to our edges
pre_sn_vertices <- c(pre_sn_codes)
pre_sn_df2 <- as.data.frame(sortproxf, stringsAsFactors = TRUE, row.names = pre_sn_vertices)
pre_sn_graph <- graph_from_data_frame(d=pre_sn_df2, vertices = pre_sn_vertices, directed = FALSE) #using the step before the matrix
#pre_sn_graph
#simplesn<-simplify(pre_sn_graph, remove.multiple = FALSE, remove.loops = TRUE)
pre_sn_proximity <- as.numeric(unlist(dplyr::select(pre_sn_combo3, "proximity"))) #Creating a numeric vector from the proximity values in order to visualize in plot
#pre_sn_proximity
#Trying to categorize by color:
pre_sn_colorrange <- colorRampPalette(c("darkred", "yellow")) #establishes color range
pre_sn_color <- pre_sn_colorrange(length(pre_sn_proximity)) #how many
#oranges <- colorRampPalette(c("dark red", "gold"))
#col <- oranges(max(pre_sn_proximity)+1)
#col <- col[pre_sn_proximity+1]
#l <- layout_on_sphere(net.bg)
#pre_sn_combo3[order(-pre_sn_combo3$proximity), ]par(bg="white")
plot(pre_sn_graph,
vertex.size=(1/25*pre_sn_proximity), #18,
vertex.color="lightgrey",
vertex.label.color="black",
edge.arrow.size=0.5, # Arrow size, defaults to 1
edge.arrow.width=0.5,
edge.width=((1/70)*(pre_sn_proximity)), #The thickness of the edges is now related to the proximity strength between indivs
edge.curved=0.25, #c(rep(0,500), rep(1,500)),
edge.color=pre_sn_color #[pre_sn_graph$proximity]
) #rep(c("darkred","yellow"), (length(pre_sn_proximity)))) #col) #"black") #pre_sn_color)
I have been able to assign the color palette, but it is not being applied properly. The thickest lines (strongest proximity value) should be the darkest color and the thinnest (weakest proximity value) should be the lightest color. I’ve been playing around with this a lot, and cannot figure out how to properly assign the color palette
2 Post-Release Proximity Plot
2.1 Inputting Datasets & Editting the dataframe
Removing Batman and possibly others: c(“RE”, “SK”, “BT”)
post_socnet<- curl("https://raw.githubusercontent.com/nickmikulski/Spring2021/main/Post-release_Social%20Proximity_CSV.csv")
post_socnet<- read.csv(post_socnet, header = T, na.strings=c(""," ","NA"))
#head(post_socnet)
#str(post_socnet)
post_socnet_close<- post_socnet %>% #creating a new dataframe called post_socnet_close using data from the original post_socnet dataframe
filter(Focal.ID != c("BT"), #this code REMOVES all data that has Batman as the focal ID (BT is wild male from prerelease)
Association != c("BT"), #this code REMOVES all data that has Batman in association column
Proximity.Code %in% c("1","2") #this code only keeps proximity codes 1,2 (excluding 3,4) because we are focusing on closer proximity
)
#head(post_socnet_close)2.2 Creating a matrix
# 1. Create a character vector of all the focal IDs in dataset:
post_sn_IDs<-sort(as.character(unique(post_socnet_close$Focal.ID)))
#post_sn_IDs
# 2. Get a list of dataframes, subsetted by monkey ID:
post_sn_monkeylist<-lapply(post_sn_IDs, function(x){post_socnet[post_socnet[["Focal.ID"]] == x, ]})
# The line above is a little bit confusing. It is creating a separate dataframe for each individual based on their focal id
#head(post_sn_monkeylist)
# 3. Group each by focal/associate, and count how many times they are observed close together:
post_sn_grouped<-
post_sn_monkeylist %>%
purrr::map(~group_by(.,Association)) %>%
purrr::map(~summarize(.,count=n()))
#post_sn_grouped
names(post_sn_grouped) <- post_sn_IDs #this gives each grouped list the name of the Focal ID
#post_sn_grouped
# 4. Set up pairwise combinations of interacting monkeys:
post_sn_monkeycombos<-list(focal=post_sn_IDs, associate=post_sn_IDs) #create list of all possible focals/associates
post_sn_filtf<- function(x, y) {x == y} #create function to filter out same-monkey pairs ("PO is close to PO")
post_sn_combo<- post_sn_monkeycombos %>% cross_df(.,.filter=post_sn_filtf) #get the filtered combined list as a dataframe
#post_sn_combo2.3 Create the dataframe and matrix
# 5. Create new dataframes with specific criteria
post_sn_combo2<-
post_sn_combo %>%
mutate(absent1 = map2_chr( #new column called "absent1"
focal,
associate,
~if_else(.x %in% names(post_sn_grouped),true="TRUE",false="FALSE"))) %>%
mutate(absent2 = map2_chr(
focal,
associate,
~if_else(.y %in% post_sn_grouped[[.x]]$Association,true="TRUE",false="FALSE"))) %>%
filter(absent1 == "TRUE") %>%
filter(absent2 == "TRUE") %>%
dplyr::select(-absent1,-absent2) #this removes those two new columns you made so you're just left with the ID names
#Honestly I was quite confused on parts of this, but Laura/Dr. Schmitt gave me this helpful code
post_sn_combo3<- post_sn_combo2 %>%
mutate(proximity = map2_int( #new column called "proximity" that is the count for when proximity code = 1 or 2
focal,
associate,
~post_sn_grouped %>% pluck(.x) %>% filter(Association==.y) %>% as.data.frame(.) %>% .[,2]))
post_sn_combo3
# 6. Create your matrix
post_sn_matrix<-spread(post_sn_combo3,associate,proximity) %>% column_to_rownames(var="focal") %>% data.matrix()
#post_sn_matrix2.5 Plotting based on dataframe
My original intention was to use the matrix that was just created in order to plot the social network, however, I keep getting an error message that says "Some vertex names in edge list are not listed in vertex data frame. I tried trouble shooting this with no success, so I instead tried to work with the dataframe that created right before the matrix (pre_sn_combo3), and I had some success. Below is how I was originally trying to create the graph using the matrix.
Alternatives (?):
post_sn_graph <- graph_from_data_frame(d=post_sn_edges, vertices = post_sn_nodes, directed = TRUE) #Using the matrix
post_sortprox <- post_sn_combo3[order(post_sn_combo3$proximity),]
#post_sortprox
post_sn_edges <- post_sn_df #assigning the matrix data frame to our edges
post_sn_vertices <- c(post_sn_codes)
post_sn_df2 <- as.data.frame(post_sortprox, stringsAsFactors = TRUE, row.names = post_sn_vertices)
post_sn_graph <- graph_from_data_frame(d=post_sn_df2, vertices = post_sn_vertices, directed = FALSE) #using the step before the matrix
#post_sn_graph
#simplesn<-simplify(post_sn_graph, remove.multiple = FALSE, remove.loops = TRUE)
post_sn_proximity <- as.numeric(unlist(dplyr::select(post_sn_combo3, "proximity"))) #Creating a numeric vector from the proximity values in order to visualize in plot
#post_sn_proximity
#Trying to categorize by color:
post_sn_colorrange <- colorRampPalette(c("darkred", "yellow")) #establishes color range
post_sn_color <- post_sn_colorrange(length(post_sn_proximity)) #how many par(bg="white")
plot(post_sn_graph,
vertex.size= 10, #(1/10*post_sn_proximity),
vertex.color="lightgrey",
vertex.label.color="black",
edge.width=((1/50)*(post_sn_proximity)), #The thickness of the edges is now related to the proximity strength between indivs
edge.curved=0.25, #c(rep(0,500), rep(1,500)),
edge.color=post_sn_color) 
#attempts at fixing the coloring
#rep(c("darkred","yellow"), (length(post_sn_proximity))))
#col)
#"black")
#post_sn_color)
#[post_sn_graph$proximity]I have been able to assign the color palette, but it is not being applied properly. The thickest lines (strongest proximity value) should be the darkest color and the thinnest (weakest proximity value) should be the lightest color. I’ve been playing around with this a lot, and cannot figure out how to properly assign the color palette
3 Pre-Release Grooming Plot
3.1 Importing the FULL pre-continuous dataset
pre_cont<- curl("https://raw.githubusercontent.com/langley1/LWTdata2016/main/2016_pre-release_cont_FULL.csv")
pre_cont<- read.csv(pre_cont, header=T, na.strings=c(""," ","NA"))3.2 Grooming Matrix+ (focal RECEIVES grooming)
# 1. Create a character vector of all the monkey IDs in your dataset:
MonkeyIDs<-as.character(unique(pre_cont$FOCAL.ID))
# 2. Get a list of dataframes, subsetted by monkey ID:
monkey.prelim<-lapply(MonkeyIDs, function(x){pre_cont[pre_cont[["FOCAL.ID"]] == x, ]})
#head(monkey.prelim)
# 3. Filter each by the behavior your want, group each by associate/recipient, and count behavior:
monkey_G.prelim<-
monkey.prelim %>%
purrr::map(~filter(.,BEHAVIOUR=="G+")) %>%
purrr::map(~group_by(.,ASSOCIATION)) %>%
purrr::map(~summarize(.,count=n()))
names(monkey_G.prelim) <- MonkeyIDs# 4. Set up your pairwise combinations of interacting monkeys:
monkeylist<-list(actor=MonkeyIDs,recipient=MonkeyIDs) #create list of all possible actors/recipients
filt <- function(x, y) {x == y} #create function to filter out same-monkey pairs ("FZ grooms FZ")
combo <- monkeylist %>% cross_df(.,.filter=filt) #get the filtered combined list as a dataframe
#head(combo)
# 5.
combo_G<-
combo %>%
mutate(absent1 = map2_chr(
recipient, # I swapped the recipient/actor **here and below
actor,
~if_else(.x %in% names(monkey_G.prelim),true="TRUE",false="FALSE"))) %>%
mutate(absent2 = map2_chr(
recipient, #**here
actor,
~if_else(.y %in% monkey_G.prelim[[.x]]$ASSOCIATION,true="TRUE",false="FALSE"))) %>%
filter(absent1 == "TRUE") %>%
filter(absent2 == "TRUE") %>%
dplyr::select(-absent1,-absent2)
G1<-combo_G %>%
mutate(grooms = map2_int(
recipient, #**here
actor,
~monkey_G.prelim %>% pluck(.x) %>% filter(ASSOCIATION==.y) %>% as.data.frame(.) %>% .[,2]))
#G1
grooming_matrix<-spread(G1,actor,grooms) %>% column_to_rownames(var="recipient") %>% data.matrix() #**here
#grooming_matrix3.2.1 Pre-Release RECEIVE Groom Plot
groom_IDs <- sort(MonkeyIDs)
pre_groom_plus_df <- as.data.frame(grooming_matrix, stringsAsFactors = TRUE)
#pre_groom_plus_df
pre_sortgroomplus <- G1[order(G1$grooms),]
#pre_sortgroomplus
pre_groomplus_edges <- pre_groom_plus_df #assigning the matrix data frame to our edges
pre_groomplus_vertices <- groom_IDs
pre_groomplus_df2 <- as.data.frame(pre_sortgroomplus, stringsAsFactors = TRUE, row.names = pre_groomplus_vertices)
pre_groomplus_graph <- graph_from_data_frame(d=pre_groomplus_df2, vertices = pre_groomplus_vertices, directed = TRUE)
#pre_groomplus_graph
pre_groomplus <- as.numeric(unlist(dplyr::select(G1, "grooms"))) #Creating a numeric vector from the grooms values in order to visualize in plot
sort_pre_groomplus<- sort(pre_groomplus)
#sort_pre_groomplus
l <- layout_with_fr(pre_groomplus_graph)
l <- norm_coords(l, ymin=-1, ymax=1, xmin=-1, xmax=1)par(bg="white")
plot(pre_groomplus_graph,
vertex.size= 8,
vertex.color="lightgrey",
vertex.label.color="black",
edge.arrow.size=0.5,
edge.width=((1/8)*(sort_pre_groomplus)),
edge.curved=0.25,
edge.color="black",
margin=0,
edge.arrow.size=0.5,
rescale=FALSE,
layout=l*1) 
3.3 Grooming Matrix- (focal GROOMS)
monkey_G2.prelim<-
monkey.prelim %>%
purrr::map(~filter(.,BEHAVIOUR=="G-")) %>%
purrr::map(~group_by(.,ASSOCIATION)) %>%
purrr::map(~summarize(.,count=n()))
names(monkey_G2.prelim) <- MonkeyIDs
# 4. Set up your pairwise combinations of interacting monkeys:
monkeylist<-list(actor=MonkeyIDs,recipient=MonkeyIDs) #create list of all possible actors/recipients
filt <- function(x, y) {x == y} #create function to filter out same-monkey pairs ("FZ grooms FZ")
combo <- monkeylist %>% cross_df(.,.filter=filt) #get the filtered combined list as a dataframe
combo_G2<-
combo %>%
mutate(absent1 = map2_chr(
actor,
recipient,
~if_else(.x %in% names(monkey_G2.prelim),true="TRUE",false="FALSE"))) %>%
mutate(absent2 = map2_chr(
actor,
recipient,
~if_else(.y %in% monkey_G2.prelim[[.x]]$ASSOCIATION,true="TRUE",false="FALSE"))) %>%
filter(absent1 == "TRUE") %>%
filter(absent2 == "TRUE") %>%
select(-absent1,-absent2)
#combo_G2
G2<-combo_G2 %>%
mutate(grooms = map2_int(
actor,
recipient,
~monkey_G2.prelim %>% pluck(.x) %>% filter(ASSOCIATION==.y) %>% as.data.frame(.) %>% .[,2]))
#G2
grooming_matrix2<-spread(G2,recipient,grooms) %>% column_to_rownames(var="actor") %>% data.matrix()
#grooming_matrix23.3.1 Pre-Release GIVE Groom Plot
groom_IDs <- sort(MonkeyIDs)
pre_groom_minus_df <- as.data.frame(grooming_matrix2, stringsAsFactors = TRUE)
#pre_groom_minus_df
pre_sortgroomminus <- G2[order(G2$grooms),]
pre_sortgroomminus
pre_groomminus_edges <- pre_groom_minus_df #assigning the matrix data frame to our edges
pre_groomminus_vertices <- groom_IDs
pre_groomminus_df2 <- as.data.frame(pre_sortgroomminus, stringsAsFactors = TRUE, row.names = pre_groomminus_vertices)
pre_groomminus_graph <- graph_from_data_frame(d=pre_groomminus_df2, vertices = pre_groomminus_vertices, directed = TRUE)
#pre_groomminus_graph
pre_groomminus <- as.numeric(unlist(dplyr::select(G2, "grooms"))) #Creating a numeric vector from the grooms values in order to visualize in plot
sort_pre_groomminus<- sort(pre_groomminus)
#Trying to categorize by color:
pre_groomminus_colorrange <- colorRampPalette(c("darkred", "yellow")) #establishes color range
pre_groomminus_color <- pre_groomminus_colorrange(length(pre_groomminus_graph)) #how many par(bg="white")
plot(pre_groomminus_graph,
vertex.size= 8,
vertex.color="lightgrey",
vertex.label.color="black",
edge.width=((1/8)*(sort_pre_groomminus)),
edge.curved=0.25,
edge.arrow.size=0.5,
edge.color="black") 
3.4 Combining Grooming Received and Grooming Given
# 1. Create a character vector of all the monkey IDs in your dataset:
MonkeyIDs<-as.character(unique(pre_cont$FOCAL.ID))
# 2. Get a list of dataframes, subsetted by monkey ID:
monkey.prelim<-lapply(MonkeyIDs, function(x){pre_cont[pre_cont[["FOCAL.ID"]] == x, ]})
#head(monkey.prelim)
# 3. Filter each by the behavior your want, group each by associate/recipient, and count behavior:
monkey_G.prelim2<-
monkey.prelim %>%
purrr::map(~filter(.,BEHAVIOUR %in% c("G+","G-"))) %>%
purrr::map(~group_by(.,ASSOCIATION)) %>%
purrr::map(~summarize(.,count=n()))
names(monkey_G.prelim2) <- MonkeyIDs
# 4. Set up your pairwise combinations of interacting monkeys:
monkeylist<-list(actor=MonkeyIDs,recipient=MonkeyIDs) #create list of all possible actors/recipients
filt <- function(x, y) {x == y} #create function to filter out same-monkey pairs ("FZ grooms FZ")
combo <- monkeylist %>% cross_df(.,.filter=filt) #get the filtered combined list as a dataframe
head(combo)## # A tibble: 6 x 2
## actor recipient
## <chr> <chr>
## 1 AL PO
## 2 KO PO
## 3 MG PO
## 4 ED PO
## 5 ZI PO
## 6 BL PO# 5.
combo_G2<-
combo %>%
mutate(absent1 = map2_chr(
actor,
recipient,
~if_else(.x %in% names(monkey_G.prelim2),true="TRUE",false="FALSE"))) %>%
mutate(absent2 = map2_chr(
actor,
recipient,
~if_else(.y %in% monkey_G.prelim2[[.x]]$ASSOCIATION,true="TRUE",false="FALSE"))) %>%
filter(absent1 == "TRUE") %>%
filter(absent2 == "TRUE") %>%
dplyr::select(-absent1,-absent2)
#combo_G2
G1_FULL<-combo_G2 %>%
mutate(grooms = map2_int( #new column called grooms
actor,
recipient,
~monkey_G.prelim2 %>% pluck(.x) %>% filter(ASSOCIATION==.y) %>% as.data.frame(.) %>% .[,2]))
G1_FULL## # A tibble: 202 x 3
## actor recipient grooms
## <chr> <chr> <int>
## 1 AL PO 1
## 2 MG PO 8
## 3 ED PO 24
## 4 BL PO 19
## 5 AU PO 12
## 6 BO PO 26
## 7 TO PO 4
## 8 AM PO 24
## 9 TI PO 10
## 10 PO AL 15
## # … with 192 more rowsgrooming_matrix_FULL<-spread(G1_FULL,recipient,grooms) %>% column_to_rownames(var="actor") %>% data.matrix()
#grooming_matrix_FULL3.4.1 Plotting Combined Grooming
3.4.2 Converting to Dataframe
groom_IDs <- sort(MonkeyIDs)
pre_groom_df <- as.data.frame(grooming_matrix_FULL, stringsAsFactors = TRUE)
#pre_groom_df3.4.3 Create graph elements & color range
pre_sortgroom <- G1_FULL[order(G1_FULL$grooms),]
#pre_sortgroom
pre_groom_edges <- pre_groom_df #assigning the matrix data frame to our edges
pre_groom_vertices <- groom_IDs
pre_groom_df2 <- as.data.frame(pre_sortgroom, stringsAsFactors = TRUE, row.names = pre_groom_vertices)
pre_groom_graph <- graph_from_data_frame(d=pre_groom_df2, vertices = pre_groom_vertices, directed = TRUE) #using the step before the matrix
#pre_groom_graph
#simplesn<-simplify(post_sn_graph, remove.multiple = FALSE, remove.loops = TRUE)
pre_grooms <- as.numeric(unlist(dplyr::select(G1_FULL, "grooms"))) #Creating a numeric vector from the grooms values in order to visualize in plot
#pre_grooms
#Trying to categorize by color:
pre_grooms_colorrange <- colorRampPalette(c("darkred", "yellow")) #establishes color range
pre_grooms_color <- pre_grooms_colorrange(length(pre_grooms)) #how many 3.4.4 Plot graph Combined
par(bg="white")
plot(pre_groom_graph,
vertex.size= 8, #(1/10*pre_grooms),
vertex.color="lightgrey",
vertex.label.color="black",
edge.width=((1/10)*(pre_grooms)), #The thickness of the edges is now related to the grooms strength between indivs
edge.curved=0.25, #c(rep(0,500), rep(1,500)),
edge.color=post_sn_color) 
3.5 Grooming Matrix+ for FEMALES ONLY (focals RECEIVES grooming)
# 1. Create a character vector of all the monkey IDs in your dataset:
pre_cont_fems<- pre_cont %>% filter(sex == "F")
MonkeyIDs_fems<-as.character(unique(pre_cont_fems$FOCAL.ID))
#MonkeyIDs_fems
# 2. Get a list of dataframes, subsetted by monkey ID:
monkey.prelim_fems<-lapply(MonkeyIDs_fems, function(x){pre_cont_fems[pre_cont_fems[["FOCAL.ID"]] == x, ]})
#head(monkey.prelim_fems)
monkey_Gfems.prelim<-
monkey.prelim_fems %>%
purrr::map(~filter(.,BEHAVIOUR=="G+")) %>%
purrr::map(~group_by(.,ASSOCIATION)) %>%
purrr::map(~summarize(.,count=n()))names(monkey_Gfems.prelim) <- MonkeyIDs_fems
#Recipient and actor have been flipped here in order to correct arrow direction
combo_Gfems<-
combo %>%
mutate(absent1 = map2_chr(
recipient,
actor,
~if_else(.x %in% names(monkey_Gfems.prelim),true="TRUE",false="FALSE"))) %>%
mutate(absent2 = map2_chr(
recipient,
actor,
~if_else(.y %in% monkey_Gfems.prelim[[.x]]$ASSOCIATION,true="TRUE",false="FALSE"))) %>%
filter(absent1 == "TRUE") %>%
filter(absent2 == "TRUE") %>%
dplyr::select(-absent1,-absent2)
G1fems<-combo_Gfems %>%
mutate(grooms = map2_int(
recipient,
actor,
~monkey_Gfems.prelim %>% pluck(.x) %>% filter(ASSOCIATION==.y) %>% as.data.frame(.) %>% .[,2]))
#G1fems
#Creating a matrix:
grooming_matrix_fems<-spread(G1fems,actor,grooms) %>% column_to_rownames(var="recipient") %>% data.matrix()
#grooming_matrix_fems
groom_fem_df <- as.data.frame(grooming_matrix_fems, stringsAsFactors = TRUE)
#groom_fem_df?graph_from_data_frame
3.5.0.1 Create graph elements & color range
pre_sortgroomf <- G1fems[order(G1fems$grooms),]
pre_sortgroomf
pre_groomf_edges <- groom_fem_df #assigning the matrix data frame to our edges
pre_groomf_vertices <- MonkeyIDs_fems
pre_groomf_df2 <- as.data.frame(G1fems, stringsAsFactors = TRUE, row.names = pre_groomf_vertices)
pre_groomf_graph <- graph_from_data_frame(d=combo_Gfems, vertices = MonkeyIDs, directed = TRUE) #using the step before the matrix
#pre_groomf_graph
pre_groomsf <- as.numeric(unlist(dplyr::select(G1fems, "grooms"))) #Creating a numeric vector from the grooms values in order to visualize in plot
#pre_groomsf
#Trying to categorize by color:
pre_groomsf_colorrange <- colorRampPalette(c("darkred", "yellow")) #establishes color range
pre_groomsf_color <- pre_groomsf_colorrange(length(pre_groomsf)) #how many 3.5.1 Plot females only graph (incomplete)
par(bg="white")
plot(pre_groomf_graph,
vertex.size= 8, #(1/10*pre_grooms),
vertex.color="lightgrey",
vertex.label.color="black",
edge.width=((1/10)*(pre_groomsf)), #The thickness of the edges is now related to the grooms strength between indivs
edge.curved=0.25, #c(rep(0,500), rep(1,500)),
edge.color=post_sn_color) 
**Check: I think this is correct because the code selected for female G+ (Grooming received), so males can give. Arrows pointing from receiving female to giving individual
4 Post-Release Grooming Plot
4.1 Importing the FULL post-continuous dataset
library(curl)
library(dplyr)
library(tidyverse)
post_cont<- curl("https://raw.githubusercontent.com/langley1/LWTdata2016/main/2016_post-release_cont_FULL.csv")
post_cont<- read.csv(post_cont, header = T, na.strings=c(""," ","NA"))
#head(post_cont)4.2 Grooming Matrix+ (focal RECEIVES grooming)
# 1. Create a character vector of all the monkey IDs in your dataset:
MonkeyIDs_postgroom<-as.character(unique(post_cont$FOCAL.ID))
# 2. Get a list of dataframes, subsetted by monkey ID:
monkey.prelim.post<-lapply(MonkeyIDs_postgroom, function(x){post_cont[post_cont[["FOCAL.ID"]] == x, ]})
#head(monkey.prelim.post)
# 3. Filter each by the behavior your want, group each by associate/recipient, and count behavior:
monkey_G.prelim.post<-
monkey.prelim.post %>%
purrr::map(~filter(.,BEHAVIOUR=="G+")) %>%
purrr::map(~group_by(.,ASSOCIATION)) %>%
purrr::map(~summarize(.,count=n()))names(monkey_G.prelim.post) <- MonkeyIDs_postgroom
# 4. Set up your pairwise combinations of interacting monkeys:
monkeylist_post<-list(actor=MonkeyIDs_postgroom,recipient=MonkeyIDs_postgroom) #create list of all possible actors/recipients
filt <- function(x, y) {x == y} #create function to filter out same-monkey pairs ("FZ grooms FZ")
combo_post <- monkeylist_post %>% cross_df(.,.filter=filt) #get the filtered combined list as a dataframe
#head(combo_post)
# 5.
combo_G_post<-
combo_post %>%
mutate(absent1 = map2_chr(
recipient,
actor,
~if_else(.x %in% names(monkey_G.prelim.post),true="TRUE",false="FALSE"))) %>%
mutate(absent2 = map2_chr(
recipient,
actor,
~if_else(.y %in% monkey_G.prelim.post[[.x]]$ASSOCIATION,true="TRUE",false="FALSE"))) %>%
filter(absent1 == "TRUE") %>%
filter(absent2 == "TRUE") %>%
dplyr::select(-absent1,-absent2)
G1_post<-combo_G_post %>%
mutate(grooms = map2_int(
recipient,
actor,
~monkey_G.prelim.post %>% pluck(.x) %>% filter(ASSOCIATION==.y) %>% as.data.frame(.) %>% .[,2]))
#G1_post
grooming_matrix_post<-spread(G1_post,actor,grooms) %>% column_to_rownames(var="recipient") %>% data.matrix()
#grooming_matrix_post4.2.1 Post-Release RECEIVE Groom Plot
groom_IDs_postrg <- sort(MonkeyIDs_postgroom)
post_groom_plus_df <- as.data.frame(grooming_matrix_post, stringsAsFactors = TRUE)
post_sortgroomplus <- G1_post[order(G1_post$grooms),]
post_groomplus_edges <- post_groom_plus_df
post_groomplus_vertices <- groom_IDs_postrg
post_groomplus_df2 <- as.data.frame(post_sortgroomplus, stringsAsFactors = TRUE, row.names = post_groomplus_vertices)
post_groomplus_df2
post_groomplus_graph <- graph_from_data_frame(d=post_groomplus_df2, vertices = post_groomplus_vertices, directed = TRUE)
post_groomplus <- as.numeric(unlist(dplyr::select(G1_post, "grooms")))
sort_post_groomplus<- sort(post_groomplus)
#Trying to categorize by color:
post_groomplus_colorrange <- colorRampPalette(c("darkred", "yellow")) #establishes color range
post_groomplus_color <- post_groomplus_colorrange(length(post_groomplus)) #how many par(bg="white")
plot(post_groomplus_graph,
vertex.size= 8,
vertex.color="lightgrey",
vertex.label.color="black",
edge.width=((1/3)*(sort_post_groomplus)),
edge.curved=0.25,
edge.color="black",
edge.arrow.size=0.5) 
4.3 Grooming Matrix- (focal GIVES grooming)
# 1. Create a character vector of all the monkey IDs in your dataset:
MonkeyIDs_postgroom<-as.character(unique(post_cont$FOCAL.ID))
# 2. Get a list of dataframes, subsetted by monkey ID:
monkey.prelim.post<-lapply(MonkeyIDs_postgroom, function(x){post_cont[post_cont[["FOCAL.ID"]] == x, ]})
#head(monkey.prelim.post)
# 3. Filter each by the behavior your want, group each by associate/recipient, and count behavior:
monkey_G.prelim.postm<-
monkey.prelim.post %>%
purrr::map(~filter(.,BEHAVIOUR=="G-")) %>%
purrr::map(~group_by(.,ASSOCIATION)) %>%
purrr::map(~summarize(.,count=n()))names(monkey_G.prelim.postm) <- MonkeyIDs_postgroom
# 4. Set up your pairwise combinations of interacting monkeys:
monkeylist_post<-list(actor=MonkeyIDs_postgroom,recipient=MonkeyIDs_postgroom) #create list of all possible actors/recipients
filt <- function(x, y) {x == y} #create function to filter out same-monkey pairs ("FZ grooms FZ")
combo_post <- monkeylist_post %>% cross_df(.,.filter=filt) #get the filtered combined list as a dataframe
#head(combo_post)
# 5.
combo_G_postm<-
combo_post %>%
mutate(absent1 = map2_chr(
actor,
recipient,
~if_else(.x %in% names(monkey_G.prelim.postm),true="TRUE",false="FALSE"))) %>%
mutate(absent2 = map2_chr(
actor,
recipient,
~if_else(.y %in% monkey_G.prelim.postm[[.x]]$ASSOCIATION,true="TRUE",false="FALSE"))) %>%
filter(absent1 == "TRUE") %>%
filter(absent2 == "TRUE") %>%
dplyr::select(-absent1,-absent2)
G1_postm<-combo_G_postm %>%
mutate(grooms = map2_int(
actor,
recipient,
~monkey_G.prelim.postm %>% pluck(.x) %>% filter(ASSOCIATION==.y) %>% as.data.frame(.) %>% .[,2]))
#G1_postm
grooming_matrix_postm<-spread(G1_postm,recipient,grooms) %>% column_to_rownames(var="actor") %>% data.matrix()
#grooming_matrix_postm4.3.1 Post-Release RECEIVE Groom Plot
groom_IDs_postgm <- sort(MonkeyIDs_postgroom)
post_groom_minus_df <- as.data.frame(grooming_matrix_postm, stringsAsFactors = TRUE)
post_sortgroomminus <- G1_postm[order(G1_postm$grooms),]
post_groomminus_edges <- post_groom_minus_df
post_groomminus_vertices <- groom_IDs_postgm
post_groomminus_df2 <- as.data.frame(post_sortgroomminus, stringsAsFactors = TRUE, row.names = post_groomminus_vertices)
post_groomminus_df2
post_groomminus_graph <- graph_from_data_frame(d=post_groomminus_df2, vertices = post_groomminus_vertices, directed = TRUE)
post_groomminus <- as.numeric(unlist(dplyr::select(G1_postm, "grooms")))
sort_post_groomminus<- sort(post_groomminus)
#Trying to categorize by color:
post_groomminus_colorrange <- colorRampPalette(c("darkred", "yellow")) #establishes color range
post_groomminus_color <- post_groomminus_colorrange(length(post_groomminus_graph)) #how many par(bg="white")
plot(post_groomminus_graph,
vertex.size= 8,
vertex.color="lightgrey",
vertex.label.color="black",
edge.width=((1/2)*(sort_post_groomminus)),
edge.curved=0.25,
edge.color="black",
edge.arrow.size=0.5) 
4.4 Combining Grooming Received and Grooming Given
# 1. Create a character vector of all the monkey IDs in your dataset:
MonkeyIDs_postgroom<-as.character(unique(post_cont$FOCAL.ID))
# 2. Get a list of dataframes, subsetted by monkey ID:
monkey.prelim.post<-lapply(MonkeyIDs_postgroom, function(x){post_cont[post_cont[["FOCAL.ID"]] == x, ]})
#head(monkey.prelim.post)
# 3. Filter each by the behavior your want, group each by associate/recipient, and count behavior:
monkey_G.prelim.post2<-
monkey.prelim.post %>%
purrr::map(~filter(.,BEHAVIOUR %in% c("G+","G-"))) %>%
purrr::map(~group_by(.,ASSOCIATION)) %>%
purrr::map(~summarize(.,count=n())) names(monkey_G.prelim.post2) <- MonkeyIDs_postgroom
# 4. Set up your pairwise combinations of interacting monkeys:
monkeylist_post<-list(actor=MonkeyIDs_postgroom,recipient=MonkeyIDs_postgroom) #create list of all possible actors/recipients
filt <- function(x, y) {x == y} #create function to filter out same-monkey pairs ("FZ grooms FZ")
combo_post <- monkeylist_post %>% cross_df(.,.filter=filt) #get the filtered combined list as a dataframe
#head(combo_post)
# 5.
combo_G_post2<-
combo_post %>%
mutate(absent1 = map2_chr(
actor,
recipient,
~if_else(.x %in% names(monkey_G.prelim.post2),true="TRUE",false="FALSE"))) %>%
mutate(absent2 = map2_chr(
actor,
recipient,
~if_else(.y %in% monkey_G.prelim.post2[[.x]]$ASSOCIATION,true="TRUE",false="FALSE"))) %>%
filter(absent1 == "TRUE") %>%
filter(absent2 == "TRUE") %>%
dplyr::select(-absent1,-absent2)
G1_post_FULL<-combo_G_post2 %>%
mutate(grooms = map2_int(
actor,
recipient,
~monkey_G.prelim.post2 %>% pluck(.x) %>% filter(ASSOCIATION==.y) %>% as.data.frame(.) %>% .[,2]))
#G1_post_FULL
grooming_matrix_post_FULL<-spread(G1_post_FULL,recipient,grooms) %>% column_to_rownames(var="actor") %>% data.matrix()
#grooming_matrix_post_FULL4.4.1 Converting to Dataframe
post_groom_df <- as.data.frame(grooming_matrix_post_FULL, stringsAsFactors = TRUE)
post_groom_df## AL AM AU BL BM BO CI ED HO KO MA NE PO RE TO
## AL NA NA 9 NA NA 2 NA NA NA 7 NA NA 1 14 9
## AM NA NA NA 20 NA 16 NA NA NA NA 8 NA 3 1 NA
## AU 23 7 NA 5 NA NA NA 17 6 15 18 NA 3 2 3
## BL NA 26 NA NA NA NA 8 6 NA 17 8 NA NA 15 NA
## BM NA 9 NA 6 NA 3 NA NA NA 2 2 NA NA NA 4
## BO NA NA 10 NA NA NA NA NA NA 4 NA NA 5 1 NA
## CI NA NA NA 12 NA NA NA NA 1 NA NA NA NA NA NA
## ED NA 40 NA 32 NA 4 NA NA NA NA 13 NA NA 15 NA
## KO NA NA 28 NA NA 8 NA NA NA NA 1 NA NA 1 NA
## MA 1 NA 10 NA 5 18 8 NA NA 2 NA NA NA 8 NA
## MG NA NA 5 NA NA 8 NA NA NA NA NA NA NA NA NA
## PO 1 NA NA NA NA 3 NA 15 NA NA NA NA NA NA NA
## RE NA NA 13 NA NA NA NA NA NA NA NA NA NA NA NA
## TI NA 7 2 NA NA NA NA NA NA 13 NA 2 NA NA 4
## TO 12 NA 15 NA 5 NA NA NA NA NA 3 NA NA NA NA4.4.2 Create graph elements & color range
post_sortgroom <- G1_post_FULL[order(G1_post_FULL$grooms),]
#post_sortgroom
postgroomid <-MonkeyIDs_postgroom
post_groom_edges <- post_groom_df #assigning the matrix data frame to our edges
post_groom_vertices <- postgroomid
post_groom_df2 <- as.data.frame(G1_post_FULL, stringsAsFactors = TRUE, row.names = post_groom_vertices) #removed sortgroom and sort(vertices)
post_groom_graph <- graph_from_data_frame(d=post_groom_df2, vertices = postgroomid, directed = TRUE) #using the step before the matrix
#post_groom_graph
post_grooms <- as.numeric(unlist(dplyr::select(G1_post_FULL, "grooms"))) #Creating a numeric vector from the grooms values in order to visualize in plot
#post_grooms
#Trying to categorize by color:
post_grooms_colorrange <- colorRampPalette(c("darkred", "yellow")) #establishes color range
post_grooms_color <- post_grooms_colorrange(length(post_grooms)) #how many 4.5 Plot graph
par(bg="white")
plot(post_groom_graph,
vertex.size= 8, #(1/10*post_grooms),
vertex.color="lightgrey",
vertex.label.color="black",
arrow.size=0.5,
edge.width=((1/5)*(post_grooms)), #The thickness of the edges is related to the grooms strength between indivs
edge.curved=0.3, #c(rep(0,500), rep(1,500)),
edge.color=post_grooms_color) 
5 Final Figures
5.1 Pre-Release Proximity Plot
5.2 Post-Release Proximity Plot
5.3 Pre-Release Grooming Plot
Batman had the lease grooming interactions with the group pre-release, which makes sense as he was the wild male who joined the group, but he was not released with the group.[]
5.4 Post-Release Grooming Plot
Zip & Jack appear to have no grooming connections post-release Neville only one connection with Tinker
6 Calculating Centrality
I found the same patterns to be true between centrality types. Either are a good representation of centrality for our purposes.
6.1 Local Centrality
6.1.1 Pre-Release Proximity Plot
degree.cent_pre_prox <- centr_degree(pre_sn_graph, mode = "all")
#degree.cent_pre_prox$res
pre_prox_lcent <- data.frame(pre_sn_IDs, (degree.cent_pre_prox$res))
#pre_prox_lcent
pre_prox_lcent[order(-pre_prox_lcent$X.degree.cent_pre_prox.res.), ]## pre_sn_IDs X.degree.cent_pre_prox.res.
## 1 PO 32
## 2 AL 32
## 3 KO 32
## 4 ED 32
## 5 ZI 32
## 6 BL 32
## 7 NE 32
## 8 AU 32
## 9 BO 32
## 10 BA 32
## 11 TO 32
## 12 MA 32
## 13 JA 32
## 14 AM 32
## 15 BM 32
## 16 TI 32
## 17 MG 32There is little/no difference in centrality within the pre-release proximity graph. This makes sense as they are in a confined location and are likely coming in contact with every other individual.
6.1.2 Post-Release Proximity Plot
degree.cent_post_prox <- centr_degree(post_sn_graph, mode = "all")
#degree.cent_post_prox$res
post_prox_lcent <- data.frame(post_sn_IDs, (degree.cent_post_prox$res))
#post_prox_lcent
post_prox_lcent[order(-post_prox_lcent$X.degree.cent_post_prox.res.), ]## post_sn_IDs X.degree.cent_post_prox.res.
## 2 AM 40
## 3 AU 40
## 1 AL 39
## 8 ED 39
## 11 KO 39
## 16 PO 39
## 4 BL 38
## 12 MA 38
## 13 MG 38
## 14 NE 37
## 20 TO 37
## 5 BM 36
## 6 BO 36
## 19 TI 36
## 10 JA 34
## 17 RE 32
## 15 PA 29
## 18 SK 27
## 7 CI 26
## 9 HO 25
## 21 ZI 23In the post-release centrality measure, we begin to see some individuals becoming more centrally important to the group. AM is the most central and ZI is the least central individual.
6.1.3 Proximity Comparison
6.1.4 Pre-Release Grooming Plot
degree.cent_pre_groom <- centr_degree(pre_groom_graph, mode = "all")
#degree.cent_pre_groom$res
pre_groom_lcent <- data.frame(MonkeyIDs, (degree.cent_pre_groom$res))
#pre_groom_lcent
pre_groom_lcent[order(-pre_groom_lcent$X.degree.cent_pre_groom.res.), ]## MonkeyIDs X.degree.cent_pre_groom.res.
## 3 KO 30
## 2 AL 29
## 16 AM 29
## 1 PO 28
## 11 BO 28
## 17 BM 27
## 7 BL 26
## 9 NE 26
## 5 ED 25
## 6 ZI 24
## 13 TO 24
## 12 BA 23
## 10 AU 19
## 15 JA 19
## 4 MG 15
## 14 MA 12
## 18 TI 12
## 8 BT 8There is a higher degree of centrality in relation to grooming compared to proximity within the pre-release data. The most central grooming partner is KO, while the least was BT. This is not surprising because BT was the wild male that joined the group before release. It makes sense that he would have the least grooming interactions.
6.1.5 Post-Release Grooming Plot
degree.cent_post_groom <- centr_degree(post_groom_graph, mode = "all")
#degree.cent_post_groom$res
post_groom_lcent <- data.frame(MonkeyIDs_postgroom, (degree.cent_post_groom$res))
#post_groom_lcent
post_groom_lcent[order(-post_groom_lcent$X.degree.cent_post_groom.res.), ]## MonkeyIDs_postgroom X.degree.cent_post_groom.res.
## 12 AU 18
## 13 MA 14
## 11 BO 12
## 3 BL 11
## 6 KO 11
## 7 AL 10
## 17 AM 10
## 15 RE 9
## 9 TO 8
## 10 BM 8
## 16 ED 8
## 4 PO 7
## 14 TI 5
## 18 CI 4
## 8 MG 2
## 19 HO 2
## 5 NE 1
## 1 ZI 0
## 2 JA 0Centrality within grooming relations is much lower and much more varied overall in the post-release data. The most central grooming partner was AU, and there was even two individuals, ZI and JA, who had no recorded grooming partners at all post-release.
6.1.6 Grooming Comparison
6.2 Global Centrality
6.2.1 Pre-Release Proximity Plot
closeness.cent_pre_prox <- closeness(pre_sn_graph, mode = "all")
#closeness.cent_pre_prox$res
pre_prox_gcent <- data.frame(pre_sn_IDs, (degree.cent_pre_prox$res))
#pre_prox_gcent
pre_prox_gcent[order(-pre_prox_gcent$X.degree.cent_pre_prox.res.), ]## pre_sn_IDs X.degree.cent_pre_prox.res.
## 1 PO 32
## 2 AL 32
## 3 KO 32
## 4 ED 32
## 5 ZI 32
## 6 BL 32
## 7 NE 32
## 8 AU 32
## 9 BO 32
## 10 BA 32
## 11 TO 32
## 12 MA 32
## 13 JA 32
## 14 AM 32
## 15 BM 32
## 16 TI 32
## 17 MG 32There is little/no difference in centrality within the pre-release proximity graph. This makes sense as they are in a confined location and are likely coming in contact with every other individual.
6.2.2 Post-Release Proximity Plot
closeness_post_prox <- closeness(post_sn_graph, mode = "all")
#closeness_post_prox$res
post_prox_gcent <- data.frame(post_sn_IDs, (degree.cent_post_prox$res))
#post_prox_gcent
post_prox_gcent[order(-post_prox_gcent$X.degree.cent_post_prox.res.), ]## post_sn_IDs X.degree.cent_post_prox.res.
## 2 AM 40
## 3 AU 40
## 1 AL 39
## 8 ED 39
## 11 KO 39
## 16 PO 39
## 4 BL 38
## 12 MA 38
## 13 MG 38
## 14 NE 37
## 20 TO 37
## 5 BM 36
## 6 BO 36
## 19 TI 36
## 10 JA 34
## 17 RE 32
## 15 PA 29
## 18 SK 27
## 7 CI 26
## 9 HO 25
## 21 ZI 23In the post-release centrality measure, we begin to see some individuals becoming more centrally important to the group. AM is the most central and ZI is the least central individual.
6.2.3 Pre-Release Grooming Plot
closeness_pre_groom <- closeness(pre_groom_graph, mode = "all")
#closeness_pre_groom$res
pre_groom_gcent <- data.frame(MonkeyIDs, (degree.cent_pre_groom$res))
#pre_groom_gcent
pre_groom_gcent[order(-pre_groom_gcent$X.degree.cent_pre_groom.res.), ]## MonkeyIDs X.degree.cent_pre_groom.res.
## 3 KO 30
## 2 AL 29
## 16 AM 29
## 1 PO 28
## 11 BO 28
## 17 BM 27
## 7 BL 26
## 9 NE 26
## 5 ED 25
## 6 ZI 24
## 13 TO 24
## 12 BA 23
## 10 AU 19
## 15 JA 19
## 4 MG 15
## 14 MA 12
## 18 TI 12
## 8 BT 8There is a higher degree of centrality in relation to grooming compared to proximity within the pre-release data. The most central grooming partner is KO, while the least was BT. This is not surprising because BT was the wild male that joined the group before release. It makes sense that he would have the least grooming interactions.
6.2.4 Post-Release Grooming Plot
closeness_post_groom <- closeness(post_groom_graph, mode = "all")## Warning in closeness(post_groom_graph, mode = "all"): At centrality.c:
## 2784 :closeness centrality is not well-defined for disconnected graphs#closeness_post_groom$res
post_groom_gcent <- data.frame(MonkeyIDs_postgroom, (degree.cent_post_groom$res))
#post_groom_gcent
post_groom_gcent[order(-post_groom_gcent$X.degree.cent_post_groom.res.), ]## MonkeyIDs_postgroom X.degree.cent_post_groom.res.
## 12 AU 18
## 13 MA 14
## 11 BO 12
## 3 BL 11
## 6 KO 11
## 7 AL 10
## 17 AM 10
## 15 RE 9
## 9 TO 8
## 10 BM 8
## 16 ED 8
## 4 PO 7
## 14 TI 5
## 18 CI 4
## 8 MG 2
## 19 HO 2
## 5 NE 1
## 1 ZI 0
## 2 JA 0Centrality within grooming relations is much lower and much more varied overall in the post-release data. The most central grooming partner was AU, and there was even two individuals, ZI and JA, who had no recorded grooming partners at all post-release.
citation(package=“igraph”)